Extent and Status of the Forests in the Ugandan Albertine Rift

47 Pages·2002·7.11 MB·English

by  Andrew Plumptre

Extent and Status of the Forests in the
Ugandan Albertine Rift
Andrew Plumptre
Wildlife Conservation Society
October 2002
Extent and Status of Forests in Ugandan Albertine Rift 2
Acknowledgements
The data presented in this report are the results of a series of biological and human impact
surveys that have taken place since the early 1990s. The biological data primarily come from
the Forest Department surveys that were supported by GEF and the European Union. The
human impact data to the forests come from surveys that the Wildlife Conservation Society
(WCS) has been undertaking in collaboration with the Jane Goodall Institute (JGI). These
surveys have been supported by the following sources to who we are grateful for their
support:
The John D. and Catherine T. MacArthur Foundation
Wildlife Conservation Society private funds
Mr. D. Thorne
Total
Nile Breweries
Crane Bank
Cleveland Parks Zoo
Berggorilla
Satellite Image analysis was undertaken in collaboration with Nadine Laporte at the
University of Maryland. This report was produced by WCS under a subcontract to the World
Wide Fund for Nature which has been contracted by the Ministry of Water, Lands and
Environment to implement the GEF PDFb Albertine Rift project in Uganda.
Wildlife Conservation Society Albertine Rift Programme
Extent and Status of Forests in Ugandan Albertine Rift 3
Executive Summary
This report summarises the current status and extent of forests in western Uganda in what is
called the Albertine Rift. The report evaluates the relative importance of the larger forest
reserves (those greater than 50 km2) for wildlife conservation and also evaluates the major
threats that the forests currently face. A satellite image analysis shows the extent of forest
cover in western Uganda based on satellite images from 1999-2002. The satellite analysis
also compares forest loss since the mid 1980s in four areas of the rift and shows that most
loss has occurred outside the forest reserves. The five forests that consistently rank high for
biodiversity conservation are: Echuya Forest Reserve, Budongo Forest Reserve, Bugoma
Forest Reserve, Kalinzu-Maramagambo Forests and Kasyoha-Kitomi Forest Reserve.
The threats faced by the forests are various but are primarily due to the following human
activities:
1. Encroachment for Agricultural land
2. Hunting for bushmeat
3. Charcoal Burning
4. Timber harvesting (where it is illegal)
5. Mining
These threats are mapped for the larger forests in western Uganda and the relative intensities
of the threats shown both between the forests and within the forest boundaries. More indirect
threats to the forest are also discussed. Supporting the ability to control the illegal activities in
these forests should be one of the main focuses of the GEF Albertine Rift project both
through support to community management initiatives and also through support to law
enforcement activities.
Wildlife Conservation Society Albertine Rift Programme
Extent and Status of Forests in Ugandan Albertine Rift 4
Table of Contents
Acknowledgements 2
Executive Summary 3
Table of Contents 4
I. Introduction 5
II Forest Cover in the Albertine Rift 7
II.1 Current forest cover 7
II.2 Forest cover change detection 9
III Conservation Value of Gazetted Forests 14
IV Threats to the Forests 16
IV.1 Major Threats 16
IV.2 Methods 16
IV.3 Mapping Human Threats 17
IV.3.1 Total Human Impact 17
IV.3.2 Hunting of Bushmeat 18
IV.3.3 Charcoal Burning 21
IV.3.4 Timber Harvesting 22
IV.3.5 Encroachment 25
IV.3.6 Mining 26
IV.4 Ranking the threats 26
IV.5. Possible Solutions to the Threats 29
V. Site Selection for the GEF Project and Development Strategies 33
V.1 Site Selection 33
V.2 Developing Conservation Strategies 34
V.2.1 Actions at national level 35
V.2.2 Actions specific to the northern corridor 36
V.2.3 Actions specific to Budongo, Bugoma, Kagombe and Itwara forest reserves 38
V.3 Conclusions 39
References 40
Appendix 1 - Gazetted Forests in the Albertine Rift 41
Appendix 2 – Description of the Larger Forest Reserves 44
Wildlife Conservation Society Albertine Rift Programme
Extent and Status of Forests in Ugandan Albertine Rift 5
Extent and Status of the Forests in the Ugandan Albertine Rift
Andrew J. Plumptre
Albertine Rift Programme,
Wildlife Conservation Society
A report under subcontract to the GEF PDFb Albertine Rift Project
implemented by WWF
I. Introduction
The Albertine Rift is an area of great importance for conservation of biodiversity. It has been
identified by Birdlife International as an Endemic Bird Area, by World Wildlife Fund as an
Ecoregion and by Conservation International as a biodiversity hotspot. It also contains some
of the richest areas in Africa for mammal and bird species. The Albertine Rift has been
defined by different people in several ways. In general it extends from the northern end of
Lake Albert down to the southern tip of Lake Tanganyika and encompasses the natural
habitats within about 100 km of the Congolese border with Uganda, Rwanda, Burundi and
Tanzania. The Uganda portion of the Albertine rift therefore extends from Budongo
Forest/Murchison Falls National Park in the north down to Mgahinga National park in the
south.
A total of 84 centrally managed forests occur in the Albertine Rift in Uganda (Appendix 1).
Five of these are national parks and 79 are central forest reserves. In addition there are 21
local forest reserves managed by the districts. Many of the forest reserves are small in size,
however, and only nine of them exceed 50 km2 in size. Figure 1.1 shows the locations of the
major forest blocks in western Uganda.
During the early 1990s the Uganda Forest Department surveyed 10 of these larger forests
(eight greater than 50 km2 and Echuya and Mafuga Forest Reserves which were deemed to be
potentially important because they occur at high altitude) and four of the national parks to
assess the relative biodiversity richness of each of them. These efforts produced species lists
for each of the forests for five taxa: trees, birds, small mammals, butterflies and moths. Not
surprisingly many of the forests in the Albertine Rift ranked high relative to other forests in
Wildlife Conservation Society Albertine Rift Programme
Extent and Status of Forests in Ugandan Albertine Rift 6
Uganda, not only for species richness but also for prevalence of restricted range or threatened
species. Data from these surveys have been used in this report.
Figure 1.1. The distribution of the larger gazetted forest areas in the Ugandan Albertine Rift.
The areas named are referred to in the text of this report.
The Wildlife Conservation Society (WCS) subsequently started a series of surveys of the
forests in collaboration with the Jane Goodall Institute (JGI) in 1999. The aim of these
surveys was to assess the distribution of larger mammal species across the forests and to
analyse the current threats to the forests and their spatial distribution within the forests. These
surveys are still ongoing but most of the large forest reserves in the Albertine Rift have been
surveyed and preliminary results are presented here.
This report provides an overview of the extent and status of the forests in the Uganda portion
of the Albertine Rift. In particular it highlights the relative importance of the larger forests for
conservation, and analyses the threats each of them faces from human activities. The report is
divided into two sections. The first section gives an analysis comparing all of the forests and
the threats impacting them, and the second section (appendix) gives a more detailed
Wildlife Conservation Society Albertine Rift Programme
Extent and Status of Forests in Ugandan Albertine Rift 7
description of the larger forest reserves which are of interest to the GEF PDFb process (ie
those forests that have not been receiving much donor support and which are not
transboundary in nature).
II. Forest Cover in the Albertine Rift
II.1 Current forest cover
Satellite images of the Albertine rift in western Uganda were analysed by Nadine Laporte,
Mirtoslav Honsak, and Didier Devers at the University of Maryland to provide an estimate of
current forest cover (using images from 1999, 2000 and 2001) and also to analyse forest loss
between the mid 1980s and the recent images. Landsat-7 images were georeferenced with
less than 1 pixel (< 30m) accuracy into the EarthSat Corporation’s GeoCover™ orthorectified
Landsat-5 image base map using a 2nd order polynomial function and the Nearest Neighbor
resampling scheme. The mosaic of the images was achieved using standard mosaicing
techniques including histogram matching and feathering of the overlap areas for a better
visual effect (Note: the original values of some of the images were significantly altered in
order to match other images acquired at different solar times, under different atmospheric
conditions and during different seasons). The images used are detailed in Table 2.1.
Table 2.1. Landsat Thematic Mapper data used to compare forest change. The dates of each
image on each satellite path are given
Data Source Landsat-5 Landsat-7
Path/Row 1980’s 1999/2002
172-59 1986-01-17 2002-02-06
172-60 1984-07-06 2001-01-02
172-61 1984-06-20 1999-07-08
173-59 1987-08-07 1999-09-17
173-60 1987-08-07 2001-12-01
173-61 1987-08-07 2000-06-15
Figure 2.1 shows the existing forest cover in western Uganda based on the analysis of the
most recent images (1999-2002). It shows that in south west Uganda the high density of
people creates a region where the boundaries of gazetted forests are hard and there is little
secondary forest/woodland in between the gazetted forests. However, in Kibaale, Hoima and
Masindi districts there are still areas where there is quite a bit of woodland and forest (often
Wildlife Conservation Society Albertine Rift Programme
Extent and Status of Forests in Ugandan Albertine Rift 8
Figure 2.1. Forest cover in western Uganda. Dark green = primary forest; Light green =
degraded/regenerating forest. The protected areas in this region are outlined in colour as
follows: National parks – red; Central forest reserves - yellow; Jointly Managed areas –
orange; Wildlife Reserves - blue and Local Forest Reserves - purple.
The image identifies forest cover well but care must be taken in the areas between Ruwenzori
and Kibale national parks and around Bwindi and Mgahinga where the forest cover is
primarily plantations rather then natural forest.
Wildlife Conservation Society Albertine Rift Programme
Extent and Status of Forests in Ugandan Albertine Rift 9
along rivers and streams) outside gazetted forest. This provides a degree of connectivity
which might be important for species dispersal and gene flow and one recommendation of
this study would be that the GEF PDFb for the Albertine Rift aims to work with local
authorities and communities to ensure some this connectivity is maintained in future.
One species for which dispersal and gene flow could be important is the chimpanzee.
Chimpanzees are classified as endangered under IUCN criteria, they occur at low densities
compared with other animals and have slow reproductive rates. Surveys WCS has carried out
with JGI in many of these forests show that most forests contain less than 500. A crude
estimate of a population that is viable in the long term is one which has at least 500
individuals animals (Soulé 1987). Hence if the populations in these forests are to remain
viable they need the connectivity between the forests to allow gene flow. Other species which
may be in a similar situation (ie they are known to occur at low densities) include large
carnivores (leopard, golden cat, servals), large ungulates (buffalos and large duikers – due to
hunting pressures many are at low density – see below) and large birds of prey (although
these can probably migrate between discontinuous forests).
II.2. Forest Cover Change Detection
Forest cover change detection was performed individually for four regions of the image
around the major forest reserves of Budongo, Bugoma, Kasyoha-Kitomi/Kalinzu and
Bwindi/Mgahinga parks. This approach was used to provide more accurate results than a
method using the entire Landsat image (or mosaic) by reducing spatially variable atmospheric
conditions and seasonal vegetation effects. The forest cover change detection consisted of
two stages: (1) classification of individual images to obtain a cloud/cloud shadow mask, (2)
classification of multi-temporal image data sets (masked for clouds/cloud shadows) in order
to obtain a forest cover change map. A combination of supervised and unsupervised
classification techniques combined with expert knowledge of how to sort clusters was used to
obtain the forest cover change maps. Both classifications were performed using three spectral
bands for each date, i.e., Landsat bands 3, 4, and 5. This combination of bands usually
contains the most useful information when mapping vegetation cover. In the first stage, a
maximum likelihood classification technique using class probabilities was used. A training
data set was obtained for the following four classes: cloud, cloud shadow, land, and open
Wildlife Conservation Society Albertine Rift Programme
Extent and Status of Forests in Ugandan Albertine Rift 10
water. The classification accuracy was enhanced by restricting the classification into an area
of interest, which was manually created as a mask prior to the supervised classification.
A spatially joined cumulative cloud/cloud shadow mask derived from both dates was
subsequently applied to each image. In the same step the corresponding images were joined
together into a 6-layer multi-temporal database. The cloud/shadow masking procedure
resulted in an underestimation of the area covered by cloud and an overestimation of cloud
shadows. These inaccuracies were usually eliminated during the second classification stage.
In the second stage, an unsupervised Isodata classification scheme was employed to cluster
the multi-temporal database. Depending on spectral variability of input data, 60 to 80 clusters
were used to capture the land cover change variability. Using this initial result and expert
knowledge of the area, each cluster was examined and assigned to one of the six classes of
landuse cover. For each region the resultant image was recoded to its respective class and
filtered (clusters of pixels of the same class smaller than four inter-connected pixels were
eliminated). Only change from “mature forest” to “non forest” was considered a forest loss or
deforestation. Land cover change from “degraded forest” to “non forest” was not considered
robust enough to be kept in the deforestation class and the resulting areas were assigned to
“non forest”. Using this conservative approach we avoid overestimating deforestation - the
rate can only be underestimated – thus the rates of conversion computed from this data set are
likely to be more reliable and robust.
Figure 2.2 shows the four regions analyse for forest loss. These show that in the southwest of
Uganda (around Bwindi and Kalinzu – Kasyoha-Kitomi), where human density is high, the
degree of forest loss around the forest blocks is low because of the low level of forest cover
here. Where forest has been lost it is primarily from plantation harvesting in this region.
Forest loss around Bugoma and Budongo is more marked and has primarily occurred outside
the protected areas on private or government owned land. Figure 2.3 shows how forest loss
varies with distance from the boundaries of the major forest blocks. It shows that forest loss
peaks between 2-4 km from the boundary for many of the forests. Budongo forest reserve is
the exception with an additional peak around 15 km from the forest edge. This is due
primarily to the development of the Kinyala sugar estate during this period which led to the
clearance of woodland and forest for sugar cane. These peaks around 2-4 km do not occur for
Bwindi and Kalinzu forests where human population density is high and where most forest
patches outside the blocks are plantations.
Wildlife Conservation Society Albertine Rift Programme